ClusterTree-RS: the binary tree algorithm for identification of co-regulated genes by clustering regulatory signals
Stavrovskaia, E.D.; Makeev, V.I.; Mironov, A.A.
Molekuliarnaia Biologiia 40(3): 524-532
2006
ISSN/ISBN: 0026-8984 PMID: 16813171 Document Number: 595259
Identification of groups of co-regulated genes (regulons) is an important part of studying transcriptional regulation. One possible approach is to cluster regulatory sites that were found using experimental or computational techniques, such as phylogenetic footprinting. This strategy doesn't require a priori knowledge about co-regulation and allows finding putative new members of known groups of co-regulated genes (i.e. a new regulon). Also, it allows finding new putative regulons, which is especially important for poorly annotated genomes. We have developed ClusterTree-RS, an algorithm for clustering regulatory signals using binary trees; it is presented in this paper along with some testing results on simulated and real data. The algorithm is implemented in Java and took about 2 hours 40 minutes to cluster 1500 input signals on a computer with AMD Athlon 1.91 GHz CPU.